s4cnnd1imema9

ADVANCES

14 Scientific American, April 2019

HEALTH TECH

Cervical

Cancer App

An AI algorithm diagnoses

the disease from images

One of the most common and cost-effec-
tive ways to detect cervical cancer is the
pap smear, in which cells are scraped from
a woman’s cervix and sent to a laboratory
for analysis. But this method requires
equipment and medical expertise that are
not always available in some low-income
countries. Now scientists are making an
app they hope could use artificial intelli-
gence to identify precancerous or cancer-
ous cells with just a photograph.
The app is being developed by research-
ers at the National Institutes of Health and

Global Good; the latter is a joint effort by Bill

Gates and invention firm Intellectual Ven-

tures. Their preliminary results, published

online in January in the Journal of the National

Cancer Institute, suggest that such an ap -

proach could significantly improve cervical

cancer diagnosis in low-resource settings.

Cervical cancer rates are higher in

countries or regions that lack the resourc-

es to conduct pap smears. Health care

providers in these areas often use a less

accurate diagnostic technique, in which

they swab the cervix with dilute acetic

acid and visually inspect it for any white

spots that might signal abnormal cells.

Over a seven-year period nih research-

ers routinely photographed the cervixes

of more than 9,400 women in Costa Rica.

They used these images to train an AI algo -

rithm to recognize characteristics of abnor-

mal tissue—and to predict later cancer

devel opment. When the algorithm analyzed

new images, it performed better than a clin-

ical expert did by visual inspection.

“We were surprised to see that comput-

ers could see much more sensitively and

clearly which cervixes are or are not pre-

cancerous,” says Mark Schiffman, a molec-

ular epidemiologist at the National Cancer

Institute and senior author of the paper. “I

really thought [the AI was] cheating.” The

scientists ultimately plan to implement their

algorithm on mobile phones and aim to

train future iterations of the program with

digital camera photos. — Wudan Yan STEVE GSCHMEISSNER

Science Source

ANIMAL FORENSICS

Catching

Paleo Killers

New technique identifies

predators of ancient animals

Nowadays detectives can use DNA analy-

sis to help catch a killer. But what happens

when a crime scene has been exposed to

the elements for thousands of years? DNA

does not always stay intact that long—so

for a paleontologist trying to figure out

what kind of predator killed a long-dead

fossil animal, the case often goes cold.

But a new method promises to help

researchers identify these ancient killers.

It relies on the fact that when a predator

gulps down the bones of its prey—say,

when a swooping owl snatches and

eats a small rodent in the night—the

diner’s stomach juices leave behind

microscopic etchings on the surface of

the victim’s bones.

These etchings occur in patterns that

are unique to the type of predator that

did the deed, making them a bit like finger-

prints that scientists can use to crack

un solved cases, explains Rebecca Terry,

a paleo ntologist at Oregon State Universi-

ty, who led the team that studied the

etchings. This technique, she adds,

will help researchers paint pic-

tures of what kinds of preda-

tors were active in long-van-

ished ecosystems, particu-

larly in areas where fossils

are scarce. “It’s really

powerful,” she says.

Terry and her team

used a scanning electron

microscope to examine

the leftover bones that

modern predatory birds

regurgitate as pellets after

a meal. They also looked at

the feces of carnivorous mam-

mals. “A bone that passes into

and out of a nocturnal owl is clearly

distinguishable from bones that have

been eaten by diurnal raptors” or mam-

mals, Terry says. Patterns etched on bones

inside an owl’s stomach tend to be relative-

ly short and close together; those from the

stomach of a hawk or mammal tend to be

longer and more widely spaced, according

to the study, which was published last

November in PALAIOS. And the patterns

left by the modern-day owls and mam-

mals, Terry adds, were “indistinguishable”

from those found on fossil bones digested

by similar predators long ago.

These findings will help answer one

of paleontologists’ most basic questions

about the fossils of animals they suspect

were killed and eaten: “Whodunit?” As

Joshua Miller, a paleobiologist at the Uni-

versity of Cincinnati, who was not involved

in the new research, says, “You can actual-

ly look at an individual bone and get some

perspective on why that bone is where you

found it. And that’s really neat.” — Lucas Joel

Cervical cancer cell

Illustration by Thomas Fuchs